MARINE ECOLOGY PROGRESS SERIESMar Ecol Prog Ser

Vol. 208: 229–248, 2000 Published December 8

INTRODUCTION

A common goal of food habits research is to predictthe composition of the diet of an animal given its size,its morphological and behavioral characteristics, andthe available food resources. Predators may feed indis-criminately on the most abundant prey in their sur-

roundings or they may feed selectively on specific preyfrom those available, with the potential for several fac-tors to contribute to the selection patterns observed(Eggers 1977, Greene 1986, Sih & Moore 1990). To thisend, the focus of a large body of empirical work hasbeen the development of quantitative relationshipsbetween patterns of prey use and attributes of bothpredator and prey (Folkvord & Hunter 1986, Hart &Hamrin 1990, Litvak & Leggett 1992, Juanes 1994,Christensen 1996).

In aquatic communities, body size of both prey andpredator is 1 attribute that has been linked directly toforaging success. Predators become more successfulwith size due to a variety of factors including increased

© Inter-Research 2000

Present addresses:**NOAA/NMFS/NEFSC, James J. Howard Marine Labora-

tory, 74 Magruder Rd., Highlands, New Jersey 07732, USA. **E-mail: fred.scharf@noaa.gov**Department of Natural Resources Conservation, University

of Massachusetts, Amherst, Massachusetts 01003, USA

Predator size - prey size relationships of marinefish predators: interspecific variation and effects ofontogeny and body size on trophic-niche breadth

Frederick S. Scharf1,*, Francis Juanes2, Rodney A. Rountree3,**

1Department of Natural Resources Conservation, University of Massachusetts, Amherst, Massachusetts 01003, USA2Department of Natural Resources Conservation and Graduate Program in Organismic and Evolutionary Biology,

University of Massachusetts, Amherst, Massachusetts 01003, USA3NOAA/NMFS/NEFSC, Woods Hole Marine Laboratory, 166 Water St., Woods Hole, Massachusetts 02543, USA

ABSTRACT: We utilized a long-term data base collected over a broad geographic range to examinepredator size - prey size relationships for 18 species of marine fish predators from continental shelfwaters off the northeast US coast. Regression analysis was used to illustrate interspecific variation inontogenetic patterns of prey size use, gape allometries, and ratio-based trophic niche breadths. Size-based feeding strategies were assessed through comparison of frequency distributions of relativeprey sizes eaten and were related to general predator feeding tactics and gape morphology. Theresults demonstrated that the range of prey sizes eaten expanded with increasing predator body sizefor each of the marine predators examined, leading to asymmetric predator size - prey size distribu-tions. Absolute maximum prey size and slopes of maximum prey size versus predator size variedwidely among predator taxa. Distinct size-based feeding strategies were evident, as diets of somepredators were dominated by prey that were 10 to 20% of predator size, whereas other predatorsfrequently consumed prey >50% of predator size. Gape sizes and allometric relationships with bodysize were also diverse among predators and often were closely associated with maximum prey sizes.Ratio-based trophic-niche breadths generally did not expand with predator ontogeny and tended tonarrow for the largest predators, which may be common for animal taxa.

KEY WORDS: Predator-prey · Body size · Piscivorous fish · Trophic niche · Gape size

Resale or republication not permitted without written consent of the publisher

Mar Ecol Prog Ser 208: 229–248, 2000

sustained and burst swimming speeds and bettervisual acuity (Keast & Webb 1966, Webb 1976,Beamish 1978, Blaxter 1986). The escape response ofprey is also strongly related to body size, as reactiondistances increase and swimming performance is en-hanced with size (Folkvord & Hunter 1986, Blaxter &Fuiman 1990). Further, the morphology of predatorsand prey change with ontogeny as predator gape sizeincreases and morphological defenses of prey (e.g.spines) become more robust. The relation betweenprey size and predator size has therefore been recog-nized as being of paramount importance in determin-ing the outcome of interactions among species.

For most fishes, the sizes of prey consumed generallyincrease with increasing predator size (Keast & Webb1966, Popova 1967, 1978, Nielsen 1980, Persson 1990,Juanes 1994). In addition, the range of prey sizes eatentypically increases in larger predators, with maximumprey size increasing while minimum prey size oftenchanges only slightly over a broad range of predatorsizes. Such asymmetric patterns in predator size - preysize distributions have generated hypotheses aboutcompetitive interactions among predators of differentsize; in terms of prey size, the diet of smaller predatorsis a subset of the diet of larger predators (Brooks &Dodson 1965, Hall et al. 1970, Wilson 1975). Thus,larger predators may have a competitive advantage byfeeding on small prey sizes while also being able tofeed on large prey that are unavailable to smallerpredators.

The prevalence of asymmetric predator size - preysize distributions in aquatic ecosystems provokesinquiry as to the nature of their existence. Why shouldlarge predators continue to include small prey in theirdiet? Traditional foraging theory predicts that animalsshould maximize their net rate of energy return whenselecting prey (Schoener 1971, Charnov 1976, Stephens& Krebs 1986), with models for particulate feedersoften predicting that the largest prey available shouldbe eaten to maximize feeding efficiency (Ivlev 1961,Werner & Hall 1974, Harper & Blake 1988). However,several examples indicate that smaller prey areselected over larger prey when predators are given achoice (Gillen et al. 1981, Hoyle & Keast 1987, Hart &Hamrin 1990, Juanes 1994), suggesting that currenciesused in foraging models to estimate the relative costsand benefits of different sized prey may be incomplete.

In attempts to characterize the ecological niche of anindividual, measures of prey size have been used torepresent the trophic niche of animal taxa (Schoener& Gorman 1968, Roughgarden 1972, 1974). Severalrecent studies have examined trophic-niche breadth infishes on a ratio scale by examining changes in thestandard deviation of log-transformed prey-size dataas predator size increases (Pearre 1986, Munk 1992,

1997, Pepin & Penney 1997). In order to identify gen-eral ontogenetic trends in patterns of prey size use byfish predators, Pearre (1986) examined several preda-tor taxa and size ranges and concluded that the rangeof relative prey sizes did not change significantly withpredator size for most species. In contrast, Pepin &Penney (1997) found significant increases in ratio-based trophic-niche breadth in 6 larval fish speciesand positive trends in 3 others. When examined collec-tively, however, the results of past studies on theontogeny of trophic-niche breadth in fishes, whetherbased on ontogenetic changes in relative or absoluteprey size, reveal several patterns that may be general-izable across taxa. Thus, a clear need remains for thefurther study of patterns of prey size use by predatorsthat are comprehensive, employing similar protocolsand incorporating several species and ontogeneticstages.

Here, we use an extensive long-term data set toexamine size-based feeding habits of a diverse groupof marine fish predators in coastal waters off the north-east US continental shelf. For the purposes of examin-ing predators that consume large prey relative to theirown size, we focus on predators that demonstrate pis-civorous feeding habits during 1 or more stages of theirlife history. We examine predator size - prey size rela-tionships in order to determine ontogenetic changes inminimum, mean, and maximum prey sizes consumed.Variation in predator size - prey size relationshipsamong different predator species are related to spe-cies-specific attributes of predators, including predatorgape size, morphology, and feeding behavior. Fre-quency distributions of relative prey sizes consumed,based on prey/predator size ratios, are comparedacross taxa to identify general size-based feedingpatterns. We also examine ontogenetic changes introphic-niche breadth of predators based on relativeprey size, rather than absolute prey size, and relateinterspecific variation in the breadth of relative preysizes to the rate of increase in maximum absolute preysize and average predator size.

MATERIALS AND METHODS

Data source. The data used for this study were col-lected as part of fisheries-independent resource moni-toring conducted in marine waters over the continentalshelf of the northeast US since 1963 by the NortheastFisheries Science Center (NMFS, NOAA). Surveydetails can be found in Azarovitz (1981). Briefly, sam-pling is conducted for approximately 8 wk during fall(September and October) and spring (March andApril) each year. In addition to fall and spring surveys,winter and summer sampling are completed during

230

Scharf et al.: Predator size - prey size relationships of fish predators

some years. Fish are captured in trawls (Yankee-typewith 10 m approximate wingspread; 12.7 cm meshwings; 1.27 cm mesh cod-end liner) equipped withrollers to make rough bottom tows possible. Trawls arefished on the bottom in waters between 3 and 200 nau-tical miles offshore from Cape Lookout, North Carolinanorthward to waters off Nova Scotia (Fig. 1).

The sampling design is stratified random, with stratadefined by latitude, depth, and historical fishing infor-mation. Approximately 350 to 400 stations are com-pleted during each season, with the number of stationsallotted to each stratum based on stratum size. Eachstation within each stratum is randomly selected andconsists of a 30 min tow at a speed of 3.5 knots. Thecatch is sorted by species and either entire samples orrepresentative subsamples of fishes are measured tothe nearest centimeter. Subsamples of predatory fishesare randomly sampled for analysis of stomach con-tents. Criteria for subsampling has changed during thesurvey period, but effort is made to insure that repre-sentative size distributions of predators are sampled ateach station or within each stratum. Samples of stom-ach contents are either processed at sea or preserved

in a 10% formalin/seawater solution and processed inthe laboratory. Prey items are enumerated, identifiedto the lowest possible taxon, weighed (or percent stom-ach volume estimated at sea), and measured to thenearest millimeter total length.

This study utilizes data collected during 1973 to 1990for which both predator size and prey size informa-tion was obtained. Although collection of prey size dataduring this period was aimed at obtaining measure-ments for fishes and commercially important inverte-brates such as squid, shrimp, and crabs, other inverte-brate prey were measured periodically throughout thesurvey period. For this study, prey include all fishesand invertebrate species that were measured.

Predators and prey. A diverse group of 18 marinefish predators that vary in size, morphology, andbehavior was selected for this study (Table 1). Each ofthe predators in the group is piscivorous throughoutsome or most of its life cycle, but each predator alsofeeds on varying amounts of invertebrate prey in addi-tion to fish prey. The group includes fishes that occupybenthic, pelagic, and demersal habitats, as well asfishes common to both inshore and offshore waters ofthis region. Represented are 9 families and 14 generathat include several flatfishes, groundfishes, sculpins,anglerfishes, fast-swimming pelagics, and a varietyof elasmobranchs. Morphological and behavioral de-scriptions of individual predator species can be foundin Bigelow & Schroeder (1953). Both juvenile and adultfishes are represented as predator sizes range from 60to 1500 mm in length, with average lengths between250 and 930 mm. The prey field encompasses a widevariety of fishes and invertebrate species that is tooexhaustive to list and describe here. Some of the mostabundant prey by weight and number consumed bythe predators listed in Table 1 include herring (familyClupeidae); sand lance (Ammodytes spp.); cod (familyGadidae); mackerel (family Scombridae); flatfish (fam-ily Pleuronectidae); pandalid shrimp (family Pandali-dae); and cancer crabs (family Cancridae). Completedescriptions of the taxonomic composition of the dietsof predators listed in Table 1 are presented for anearlier time period in Edwards & Bowman (1979),Langton & Bowman (1980, 1981), and Sedberry (1983).

Data analysis. Interspecific variability and changesin prey size use patterns with ontogeny were exam-ined using regression analysis. Prey size - predator sizescatter diagrams were plotted for each predator spe-cies and least-squares regressions were generated toestimate the relationship between mean prey size andpredator size. To estimate changes in minimum andmaximum prey size with increasing predator size, theedges of each scatter diagram were analyzed usingquantile regression procedures (Scharf et al. 1998b,Cade et al. 1999). Regression quantiles between 0 and

231

Fig. 1. Map of northeast US coast from Cape Hatteras, NorthCarolina (S) to Nova Scotia, Canada (N). Continuous lines =sampling strata (depth zones) for National Marine FisheriesService bottom-trawl surveys conducted in continental shelfwaters; outermost continuous line = depths >185 m; singledashed line on right: position of USA/Canadian border

(Hague line)

Mar Ecol Prog Ser 208: 229–248, 2000

100 can be estimated through minimization of the sumsof the absolute values of residuals. The estimatedquantiles are generally robust to outlying values alongthe y-axis and the procedure is ideally suited to theexamination of data with heteroscedastic error distrib-utions (StataCorp. 1999, Scharf et al. 1998b). Quantilesranged between the 90th and 99th for upper boundsand the 1st and 10th for lower bounds, and were basedon sample sizes for individual predators (Scharf et al.1998b, Cade et al. 1999). Correlation analyses wereperformed among minimum, mean, and maximumprey size slope estimates to determine whether in-creases in mean prey size were primarily a result ofchanges in minimum or maximum prey size. Patternsof relative prey size use among predators were exam-ined by generating relative frequency histograms ofprey size/predator size ratios consumed by eachpredator.

As part of fisheries monitoring surveys conducted in1995 and 1996, predator species were collected toobtain morphological measurements of gape heightand gape width. Sufficient numbers and size rangeswere collected for 8 of the predator species listed inTable 1. Gape height was defined as the maximum lin-ear distance between the upper and lower jaws withthe mouth stretched open. Gape width was defined asthe linear distance between the left and right cornersof the open mouth. Least-squares regressions wereperformed to determine gape size versus body sizerelationships for the 8 predators. Lengths of prey eatenby these predators were converted to dorsoventralbody depths using depth/length relationships specific

to each prey taxon (Scharf et al. 1998c). For each scat-ter diagram of prey body depth versus predator length,maximum prey body depths consumed were estimatedusing regression quantiles. Slopes of prey body depthversus predator size regressions were compared togape estimates using post-hoc tests (Sokal & Rohlf1995, p. 500).

Trophic-niche breadth was examined on a ratio scaleby determining changes in the range of relative preysizes with increasing predator size. Prey size datawere converted to a ratio scale by dividing each preysize by its corresponding predator size and then plot-ting these prey/predator size ratios versus predatorsize. Regression quantiles (90th and 10th) were gener-ated to estimate the extremes of the ratio scale data foreach predator. Using regression quantiles to estimatethe rate of change in the extremes of the data elimi-nated the need to subjectively partition predator sizedata into separate size classes for the purposes of esti-mating means and standard deviations within eachsize class, as has been done in previous ratio-basedtrophic-niche breadth studies (Pearre 1986, Munk1992, 1997, Pepin & Penney 1997). Slope comparisonswere made between upper and lower bounds, withsignificant differences indicating an increase or de-crease in ratio-based trophic-niche breadth with in-creasing predator size. Convergent slopes indicated adecrease; divergent slopes indicated an increase. Thedifference between predicted values of upper andlower bound regressions at any given predator sizerepresented the trophic-niche breadth. For each pre-dator species, predicted niche breadths were re-

232

Predator species Mean size Range n

Spotted hake Urophycis regia 250 0070–400 0286Fourspot flounder Paralichthys oblongus 280 0160–390 0450Windowpane Scopthalmus aquosus 280 0170–390 0307Longhorn sculpin Myoxocephalus octodecimspinosus 280 0120–370 0299Sea raven Hemitripterus americanus 310 0100–620 0403Silver hake Merluccius bilinearis 320 0060–620 1391Red hake Urophysis chuss 350 0110–630 0552Weakfish Cynoscion regalis 350 0110–790 0523Summer flounder Paralichthys dentatus 360 0190–820 0373Little skate Raja erinacea 430 0110–540 0838Bluefish Pomatomus saltatrix 470 0110–880 0991White hake Urophycis tenuis 520 00150–1280 0791Goosefish Lophius americanus 600 00130–1120 0409Atlantic cod Gadus morhua 670 00210–1500 2346Pollock Pollachius virens 700 00270–1060 0271Winter skate Raja ocellata 760 00290–1090 1290Spiny dogfish Squalus acanthius 820 00230–1100 3253Smooth dogfish Mustelus canis 930 00460–1250 1313

Table 1. Marine fish predators from US northeast continental shelf waters examined in this study, listed in ascending order ofmean size. Mean size and range are total length in mm; n = number of individual prey fishes with length information recovered

from the stomachs of each predator species

Scharf et al.: Predator size - prey size relationships of fish predators

gressed against predator size and the resultant slopeswere compared to the slopes for maximum and mini-mum prey size generated from the original scatterdiagrams. This allowed us to determine whether ornot predators that increased their maximum or mini-mum prey size rapidly as they grew experienced anincrease or decrease in trophic-niche breadth relativeto those predators that increased their maximum orminimum prey size more slowly. Average trophic-niche breadth was also compared to average predatorsize to evaluate whether a general size-based trendwas apparent.

RESULTS

The range of absolute prey sizes eaten expandedwith increasing body size for all predators examined.As body size increased, all predators showed a signifi-cant increase in maximum prey size consumed, while12 of 18 predators increased minimum prey sizes eaten(Fig. 2, Table 2). Upper-bound slopes spanned an orderof magnitude, ranging from 0.077 to 0.894 (Fig. 3,Table 2). Although ranging from 0.0 to 0.170, lower-bound slopes were less variable among predators, asall but 4 slopes were <0.030 (Fig. 3, Table 2). The con-tinued inclusion of small prey in the diet across a rangeof predator body sizes resulted in asymmetric predatorsize - prey size distributions for each predator. Meanprey size - predator size slopes were positively related

to upper bound slopes (r = 0.802; p < 0.001) and werenot related to lower bound slopes (r = 0.308; p = 0.213),demonstrating that ontogenetic changes in mean preysizes eaten were driven primarily by changes in maxi-mum prey sizes consumed. Upper- and lower-boundslope estimates were not correlated (r = –0.026; p =0.918), indicating that predators with rapid ontoge-netic increases in maximum prey size did not necessar-ily increase minimum prey size rapidly.

A considerable amount of variation was presentamong predators in the relative frequency distribu-tions of prey size/predator size ratios (Fig. 4). Severalpredators consumed mainly prey that were small frac-tions of their body size. For example, nearly 75% of thediet of the predators illustrated in Fig. 4a consisted ofprey <20% of their own body size. Two of these preda-tors, smooth dogfish and little skate, had diets thatwere dominated by invertebrates, resulting in 90% oftheir diets being made up of small prey (<20% preda-tor size). On the other hand, the predators depicted inFig. 4b incorporated a considerable number of inter-mediate-sized prey in their diets while they also con-tinued to feed on several small prey. For this group ofpredators, 30 to 40% of their diets consisted of prey>20% of their own body size. The predators illustratedin Fig. 4c consumed an abundance of relatively largeprey, as 45 to 75% of the diet was made up of prey>20% of their own body size, including several preygreater than 50% of their body size. For example,nearly 30% of the diets of silver hake and white hake

233

Predator species Maximum Mean Minimum n Quantiles

Spotted hake PY = 0.495PD + 5.579 PY = 0.227PD – 1.544 PY = 0.020PD + 4.800NS 286 90th/10thFourspot flounder PY = 0.367PD – 15.667 PY = 0.218PD – 24.641 PY = 0.011PD + 2.111NS 450 95th/5thWindowpane PY = 0.419PD – 1.563 PY = 0.395PD – 52.123 PY = 0.170PD – 31.000 307 95th/5thLonghorn sculpin PY = 0.330PD + 15.900 PY = 0.006PD + 37.154NS PY = 0.021PD + 0.929NS 299 95th/5thSea raven PY = 0.894PD + 22.000 PY = 0.512PD – 13.819 PY = 0.005PD + 18.524NS 403 95th/5thSilver hake PY = 0.580PD + 27.400 PY = 0.416PD – 37.106 PY = 0.027PD – 0.577 13910 99th/1stRed hake PY = 0.633PD – 58.000 PY = 0.270PD – 32.517 PY = 0.011PD + 4.643 552 95th/5thWeakfish PY = 0.156PD + 60.976 PY = 0.060PD + 36.320 PY = 0.053PD + 1.842 523 95th/5thSummer flounder PY = 0.333PD – 3.333 PY = 0.209PD – 6.395 PY = 0.100PD – 16.000 373 90th/10thLittle skate PY = 0.320PD + 0.800 PY = 0.075PD + 7.981 PY = 0.022PD – 3.000 838 95th/5thBluefish PY = 0.462PD + 30.769 PY = 0.112PD + 34.894 PY = 0.016PD + 5.920 991 99th/1stWhite hake PY = 0.471PD + 48.824 PY = 0.268PD + 14.989 PY = 0.089PD – 20.395 791 95th/5thGoosefish PY = 0.323PD + 185.077 PY = 0.129PD + 120.320 PY = 0.002PD + 33.293NS 409 95th/5thAtlantic cod PY = 0.435PD + 32.609 PY = 0.199PD – 21.107 PY = 0.022PD – 6.741 23460 99th/1stPollock PY = 0.259PD + 51.852 PY = 0.135PD + 30.917 PY = 0.000PD + 30.000NS 271 95th/5thWinter skate PY = 0.133PD + 136.667 PY = 0.130PD + 8.561 PY = 0.015PD – 1.692 12900 99th/1stSpiny dogfish PY = 0.324PD + 70.588 PY = 0.112PD + 23.072 PY = 0.014PD + 0.864 32530 99th/1stSmooth dogfish PY = 0.077PD + 73.077 PY = 0.049PD + 12.758 PY = 0.015PD + 1.846 13130 95th/5th

Table 2. Regression equations relating maximum, mean, and minimum prey size to predator size for marine fish predators inTable 1. Estimates for maximum and minimum prey sizes were generated using quantile regression techniques; estimates formean prey sizes are least-squares estimates. PY = maximum, mean, or minimum prey total length in millimeters; PD = predatortotal length in mm; quantiles listed are those used to estimate upper and lower bounds of each prey size/predator size scatter plot.

NS = regression coefficient not significant; p > 0.05

Mar Ecol Prog Ser 208: 229–248, 2000

were made up of prey >40% of their size, whereasover 30 and 45% of the diets of goosefish and searaven, respectively, consisted of prey >50% of preda-tor size. The diets of several members of this group(bluefish, white hake, sea raven, goosefish) were moreconcentrated on fish, rather than invertebrates, when

compared to other predators (Fig. 2). For the 8 preda-tors with gape size data, ratios of gape size relative topredator size ranged from 0.04 for spiny dogfish to 0.24for goosefish and were strongly correlated with theproportion of prey in the diet that was >40% of preda-tor size (r = 0.887; p = 0.003).

234

Fig. 2. Predator size - prey size scatter diagrams for predatory fishes from marine waters off US northeast continental shelf. Eachsymbol represents a single prey consumed by a predator. (S) Fish prey; (d) invertebrate prey; regression lines = ontogeneticchanges in prey sizes consumed with increasing predator size for all prey combined; dashed lines = minimum and maximum preysizes; continuous lines = mean prey sizes. Regression equations are presented in Table 2. Predators are shown in ascending order

of mean size from upper left to lower right and from (a) to (c)

Scharf et al.: Predator size - prey size relationships of fish predators

Gape sizes of predators increased linearly withincreasing body size, however, regression slopes werevaried (Fig. 5, Table 3). Spiny dogfish and red hakeshowed only gradual increases in gape size withincreasing body size, whereas gape sizes of sea ravenand goosefish increased nearly 4 times faster throughontogeny. For 5 of 8 predators examined, gape sizeappeared to be a good indicator of maximum prey sizeconsumed, as gape size regressions had statisticallysimilar slopes to upper-bound prey size regressions

(Table 3). Elevations (y-intercepts) of gape size regres-sions were higher than those of upper-bound prey sizeregressions; however, several large prey were locatedabove the upper-bound prey size regression for eachpredator. For 3 predators (goosefish, winter skate, andbluefish), upper-bound prey size regressions weremuch lower than gape size regressions, and slopeswere significantly different. Although gape height andwidth measurements generally scaled similarly withincreasing body size, gape height more closely corre-

235

Fig. 2 (continued)

Mar Ecol Prog Ser 208: 229–248, 2000

sponded with maximum prey size eaten for 4 of 5predators based on the slopes and significance levelsof comparison tests presented in Table 3.

Only 7 of the 18 predators examined demonstrated asignificant change in ratio-based trophic-niche bread-th with increasing body size (Table 4). The generaltrend was for smaller average-sized predators to showno significant change, while the largest predatorstended to show a decrease in ratio-based trophic-nichebreadth with increasing body size (Fig. 6). For exam-

ple, 5 of the 7 largest average-sized predators haddecreasing ratio-based trophic-niche breadths, where-as the 6 smallest average-sized predators demon-strated no ontogenetic change in ratio-based trophic-niche breadth (Table 4). Although variable, averageratio-based trophic-niche breadth demonstrated adeclining trend with increasing average predator sizeacross the range of predators examined (Fig. 7). Slopesgenerated from regressions of the difference betweenupper and lower bounds of relative prey size data ver-

236

Fig. 2 (continued)

Scharf et al.: Predator size - prey size relationships of fish predators

sus predator size were positively related to the slopesof maximum prey size versus predator size regressionspresented in Table 2 and were not significantly relatedto the slopes of minimum prey size versus predator sizeregressions (Fig. 8).

DISCUSSION

Predator size - prey size distributions were asymmet-ric, with the range of prey sizes eaten expanding withincreasing predator body size for each of the marine

predators examined. Despite the consistency in thegeneral shapes of the distributions, maximum prey sizeand the rate of change in maximum prey size withontogeny varied widely among predator taxa. Differ-ential size-based feeding strategies were evident, aspredator diets were concentrated on varying ranges ofrelative prey sizes from prey that were 10 to 20% ofpredator size to prey >50% of predator size. Similar tothe variability present in prey sizes consumed, gapesizes and allometric relationships with body size werediverse among predators. Lastly, ratio-based trophic-niche breadths generally did not expand with predatorontogeny and tended to narrow for the largest preda-tors.

Existence of asymmetric predator size - prey size patterns

The general patterns of prey size use by predators inthis study correspond well with previous studies ofpredation by fishes (Popova 1967, 1978, Shirota 1970,Hunter 1981, Persson 1990, Juanes & Conover 1995,Mittelbach & Persson 1998). Collectively, these find-ings support the notion of a widespread existence ofasymmetric predator size - prey size distributions inaquatic ecosystems. The continued inclusion of smallprey in the diets of larger predators contrasts with pre-dictions of optimal foraging models for particulatefeeders, which indicate that the largest prey availableshould be consumed preferentially to maximize netenergetic return (Ivlev 1961, Werner & Hall 1974,Gillen et al. 1981, Harper & Blake 1988). However,many previous foraging models have used only thetime required to handle individual prey to representforaging costs (Werner 1974, Gillen et al. 1981, Stein etal. 1984), ignoring potential prey behavior that canaffect encounter rates and capture efficiencies (Sih &Moore 1990). For planktivorous fishes, models thatestimate foraging costs based only on prey handlingtime have proven successful in predicting patterns ofprey selection (Werner & Hall 1974, Kislalioglu & Gib-son 1976, Mittelbach 1981, Wanzenböck 1995). Pisci-vores, however, generally spend less time feedingcompared to planktivores (Breck 1993). Therefore,handling time may be of little importance relative tosize-dependent capture success and differentialencounter probabilities in determining patterns of preyselection by piscivorous fishes (Breck 1993, Juanes &Conover 1994).

Prey vulnerability is strongly affected by body size,and can be expressed as the product of encounterprobability and susceptibility to capture (Greene 1983,1986, Juanes 1994). Predator capture success is knownto decrease monotonically with increasing prey size for

237

Fig. 3. Regressions illustrating ontogenetic changes in maxi-mum, mean, and minimum prey sizes consumed with increas-ing predator size for all marine fish predators examined in thisstudy. Regression equations are presented in Table 2. Eachgraph is intended to illustrate among-predator variation inprey sizes eaten. As reference points, positions of several, but

not all, predators are indicated

Mar Ecol Prog Ser 208: 229–248, 2000

a wide variety of fish taxa (Folkvord & Hunter 1986,Miller et al. 1988, Rice et al. 1993, Juanes & Conover1994, Ellis & Gibson 1997, Scharf et al. 1998a). Much

less is known about the size-dependence of encounterrates. Based on increases in predator reactive dis-tances and prey swimming speeds with increasing

238

Fig. 4. Relative frequency distributions of prey size/predator size ratios consumed by marine fish predators. Cumulative fre-quencies are indicated by continuous lines with filled circles at 5% intervals of prey size/predator size ratios. (a) Predators thatconsume highest percentages of small prey relative to their body size (~75% or more of prey < 0.20 prey size/predator size; ~2%or less of prey > 0.40 prey size/predator size); (b) predators consuming mostly intermediate-sized prey (~50 to 75% of prey < 0.20prey size/predator size; ~2 to 7% of prey > 0.40 prey size/predator size); (c) predators that consume the highest percentages oflarge prey relative to their body size (~50% or less of prey < 0.20 prey size/predator size; ~8 to 54% of prey > 0.40 prey

size/predator size). For predators with gape information, average ratio of gape size to body size (Gape/TL) is indicated

Scharf et al.: Predator size - prey size relationships of fish predators

prey size, the probability of encounter is generallythought to increase with prey size, especially atsmaller body sizes (Gerritsen & Strickler 1977). Thecombination of increasing encounter probability anddecreasing susceptibility to capture with individualbody size produces a dome-shaped vulnerability curvefor individual prey, with intermediate prey sizes beingmost vulnerable to predation. However, although theprobability of an individual prey being encountered bya predator may increase with increasing body size,theory predicts that the total biomass of particles inmarine ecosystems declines with increasing particlesize (Sheldon et al. 1972, Kerr 1974, Platt & Denman1978). Therefore, the numbers of smaller prey avail-able to predators may be substantially greater than thenumbers of larger prey. High relative abundance ofsmall prey may outweigh any size-based increases inindividual encounter probabilities for larger prey,resulting in high predator encounter rates with smallprey relative to large prey. The combination of highencounter rates, due to greater relative abundance,and high capture probability, means that small preyshould always remain highly vulnerable to predationrelative to larger prey. Considering that handling timesincrease rapidly with prey size, the retention of smallprey in the diets of larger predators may reflect prof-itable foraging decisions by predators when search,capture, and handling costs are each considered.

Although predator size - prey size distributions foreach predator species maintained the same general

shape, considerable variability existed among preda-tors in the sizes of prey consumed and ontogeneticrelationships. In our study, increases in mean prey sizewere strongly correlated with increases in maximumprey size and were not statistically related to changesin minimum prey size. This is likely to be a commontheme in patterns of prey size use because ontogeneticchanges in maximum prey size will determine theoverall shape of the distribution while minimum preysize remains relatively constant.

Influence of predator morphology and behavior

Observed differences among predators in patterns ofprey size use can be attributed to variability in severalmorphological and behavioral characteristics. Differ-ence in mouth morphology alone has been implicatedas a major determinant of variation in prey types andsizes consumed by predatory fishes (Keast & Webb1966, Shirota 1970, Hunter 1981, Persson et al. 1996,Labropoulou & Eleftheriou 1997, Piet et al. 1998). Gapesize varied nearly 4-fold among the predators exam-ined in this study, with differences in gape size corre-sponding well with differences in the distribution ofrelative prey sizes eaten by each predator. For exam-ple, gape size to predator size ratios ranged from 0.04to 0.24 and coincided directly with the percentages oflarge prey included in the diet (prey >40% predatorbody size), which ranged from 1–2% for small-gaped

239

Fig. 4 (continued)

Mar Ecol Prog Ser 208: 229–248, 2000

predators to 40–50% for large-gaped predators.Extremely large gape sizes allow predators such as searaven and goosefish to exploit prey up to 80–90% oftheir own body size — resources normally utilized bypredators twice their size. Predators that increase gape

size rapidly also tend to feed primarily on fish preyrelative to invertebrates. For example, sea raven andgoosefish each have diets concentrated on fishes,including several larger species of clupeids andgadids. In contrast, smaller-gaped predators such as

240

Fig. 5. Predator length - prey body depth scatter diagrams for 8 piscivorous fish predators. Each symbol (S) represents a singleprey consumed by a predator. For each scatter diagram, thick continuous lines represent gape width vs predator size regressionsand thick dashed lines represent gape height vs predator size regressions. Regression equations are presented in Table 3. Maxi-mum prey body depth vs predator size regressions are also illustrated (thin continuous lines) with corresponding quantiles

indicated. Only gape width information was available for bluefish

Scharf et al.: Predator size - prey size relationships of fish predators 241

Predator species Gape size Gape at 400 mm r2 Size range n p-value

Winter skate GW = 0.108PD – 15.941 27 0.90 400–790 042 <0.001GH = 0.117PD – 19.699 27 0.87 400–790 042 <0.001PYBD = 0.012PD + 8.446

Spiny dogfish GW = 0.066PD – 1.948 24 0.69 180–840 101 >0.250GH = 0.051PD + 11.230 32 0.54 180–840 100 >0.500PYBD = 0.056PD + 2.668

Red hake GW = 0.064PD + 4.609 30 0.48 205–398 011 >0.750GH = 0.043PD + 16.905 34 0.41 205–398 011 <0.050PYBD = 0.094PD – 14.541

Atlantic cod GW = 0.128PD – 11.689 40 0.91 300–910 018 <0.050GH = 0.085PD + 1.584 36 0.87 300–910 018 >0.750PYBD = 0.090PD – 8.911

Silver hake GW = 0.076PD + 1.011 31 0.90 100–400 083 <0.001GH = 0.123PD + 2.147 51 0.81 100–400 083 >0.250PYBD = 0.134PD – 11.154

Sea raven GW = 0.244PD – 12.095 86 0.95 120–580 043 < 0.001GH = 0.170PD + 6.217 74 0.60 120–580 042 >0.750PYBD = 0.166PD – 14.441

Goosefish GW = 0.202PD + 15.881 97 0.81 160–580 009 <0.001GH = 0.242PD + 0.008 97 0.93 160–580 009 <0.001PYBD = 0.026PD + 37.596

Bluefish GW = 0.129PD + 0.840 53 0.96 90–260 021 <0.001PYBD = 0.030PD + 30.284

Table 3. Linear regression equations relating gape width, gape height, and maximum prey body depth consumed to total bodylength for 8 predator species. Only gape width was measured for bluefish. GW = gape width; GH = gape height; PYBD = preybody depth; PD = predator total length. All measurements and size range are in mm total length. r2 values, size range, and n referto gape equations only; regression p-values = < 0.05 for all gape equations. n for body-depth equations are indicated in Fig. 5.Probability values presented are for post-hoc slope comparisons between each gape equation and equation for maximum prey

body depth eaten

Predator species Prey size/predator size ratio vs predator size Change in trophic-Upper bound slope (±SE) Lower bound slope (±SE) niche breadth

Spotted hake –0.000120 (±0.000871) –0.000072 (±0.000077) NoneFourspot flounder 0.000563 (±0.000314) 0.000012 (±0.000038) NoneWindowpane 0.000545 (±0.000389) 0.000468 (±0.000158) NoneLonghorn sculpin –0.000680 (±0.000293) –0.000120 (±0.000053) NoneSea raven 0.000132 (±0.000282) –0.000167 (±0.000059) NoneSilver hake 0.000254 (±0.000084) 0.000280 (±0.000037) NoneRed hake 0.000675 (±0.000217) –0.000059 (±0.000006) IncreaseWeakfish –0.000422 (±0.000110) –0.000124 (±0.000039) DecreaseSummer flounder 0.000013 (±0.000128) 0.000137 (±0.000039) NoneLittle skate –0.000022 (±0.000226) 0.000017 (±0.000011) NoneBluefish –0.000133 (±0.000061) –0.000194 (±0.000014) NoneWhite hake –0.000228 (±0.000040) 0.000107 (±0.000035) DecreaseGoosefish –0.000770 (±0.000097) –0.000269 (±0.000042) DecreaseAtlantic cod –0.000074 (±0.000027) 0.000042 (±0.000005) DecreasePollock –0.000087 (±0.000046) –0.000106 (±0.000015) NoneWinter skate –0.000189 (±0.000016) 0.000088 (±0.000012) DecreaseSpiny dogfish –0.000064 (±0.000027) 0.000009 (±0.000006) DecreaseSmooth dogfish –0.000019 (±0.000031) –0.000004 (±0.000003) None

Table 4. Change in ratio-based trophic-niche breadth with increasing predator size for marine fishes on the US northeast conti-nental shelf. Upper- and lower-bound slopes are quantile regressions estimating 90th (upper) and 10th (lower) quantiles ofprey/predator size ratio vs predator size plots. Decreases or increases in trophic-niche width are based on statistically significant

differences between upper- and lower-bound slope

Mar Ecol Prog Ser 208: 229–248, 2000

spiny dogfish and Atlantic cod consume considerablymore invertebrates, although larger prey eaten aremainly fishes (Fig. 2c). For both marine and freshwaterfishes, previous studies have concluded that largegape relative to body size leads to an earlier shift to apiscivorous diet and that the timing of ontogenetic diet

shifts have important effects on growth during the firstyear of life (Buijse & Houthuijzen 1992, Juanes &Conover 1995, Buckel et al. 1998, Mittelbach & Persson1998). Differences in gape allometries in later lifestages, as observed for the predators in this study, canalso impact diet composition and may have conse-

242

Fig. 6. Scatter diagrams illustrating relative prey size (prey size/predator size ratio) as a function of predator size. Each symbol (S)represents relative size of a single prey consumed by a predator. Regression lines illustrate ontogenetic changes in the range ofrelative prey sizes consumed with increasing predator size; regression slopes for upper (90th quantile) and lower (10th quantile)bounds are presented in Table 4. Predators are shown in ascending order of mean size from upper left to lower right and from

(a) to (c)

Scharf et al.: Predator size - prey size relationships of fish predators

quences for growth and survival when resources arelimited.

Gape size is not the only factor limiting maximumprey size consumed by fish predators. For example,Gaughan & Potter (1997) compared the diets andmouth sizes of 5 estuarine species of larval fishes andconcluded that mouth width had only a small influenceon prey sizes eaten and that disparate feeding patternsamong larvae were likely to be due to behavioral dif-

ferences. Other studies have also observed maximumprey sizes for fish predators well below morphologicallimits set by gape size, and have suggested that size-dependent prey-evasive behaviors and differencesin prey availability may also limit the consumptionof large prey (Hambright 1991, Keeley & Grant 1997,Brooking et al. 1998). In our study, maximum preysizes for goosefish, winter skate, and bluefish werealso below those predicted by gape size. Our findings

243

Fig. 6 (continued)

Mar Ecol Prog Ser 208: 229–248, 2000

for these 3 predators provide further evidence for thepotential importance of factors other than predatorgape size, such as behavior, prey availability, andpredator capture efficiency, in limiting maximum preysize.

Size-based feeding strategies may also be related topredator foraging tactics, habitat overlap with prey,and morphological specializations that are particularlysuited to specific habitats and/or prey types. Mostpredators utilize 1 of 2 basic encounter tactics, either

lying in wait to ambush prey or cruising continually tolocate prey (Greene 1986). Because of the dependenceon prey movement to produce encounters, ambushpredators are thought to encounter large, fast-movingprey more frequently than small, slow-moving prey(Greene 1983, 1985). The predators included in thisstudy that exclusively employ ambush tactics, goose-fish and sea raven, consumed the largest prey relativeto their own size compared with other predators, whichis consistent with theoretical predictions. Several pre-

244

Fig. 6 (continued)

Scharf et al.: Predator size - prey size relationships of fish predators

dators in this study are bottom-oriented, and diets ofthese species may be biased towards the types andsizes of prey occupying benthic habitats. For example,the diets of several flatfishes, demersal groundfish, andelasmobranchs consist of an abundance of benthicprey including invertebrate crustaceans as well as fishspecies such as sand lance. Moreover, several elasmo-branch predators included in this study have ventrallylocated mouths, with the feeding apparatus of skatesbeing specialized to crush hard-bodied prey. Togetherwith a benthic lifestyle, mouth morphology and loca-tion probably have significant effects on patterns ofprey type and size use for these predators. In contrast,predators with a more pelagic lifestyle, such as blue-fish, feed mainly on schooling fishes and squid that arelocated up in the water column. In addition to mor-phology, general feeding tactics, and habitat associa-

tions, seasonal movements of predators related to bothlatitude and depth result in varying patterns of spatialand temporal overlap with prey, which will ultimatelycontribute to patterns of prey size use.

Trophic-niche breadth variation as a function ofrelative prey size

Ontogenetic changes in ratio-based trophic-nichebreadths were not evident for many of the predatortaxa we examined, which is consistent with severalprevious studies (Pearre 1986, Munk 1992, 1997). Formost of the smallest predators, we observed no signifi-cant change in trophic-niche breadths. For several ofthe largest predators, however, there was a generaltrend to indicate a decrease in the breadth of relativeprey sizes eaten with ontogeny. Pearre (1986) con-ducted the most inclusive study of trophic-nichebreadths in fishes, examining 45 data sets on 43 spe-cies. Although broad taxonomically, the species in-cluded were mainly larvae and small juveniles, withonly 3 species larger than 200 mm average size and anadditional 6 species larger than 100 mm average size(Pearre 1986). Other recent studies of trophic-nichebreadth have similarly been focused on larval andjuvenile fishes (Munk 1992, 1997, Pepin & Penney1997).

The trend for decreasing trophic-niche breadthswith ontogeny observed in our study was only presentin large predators with average sizes >500 mm. Thisfinding is likely to be a common one when large preda-tors are examined for ontogenetic changes in the rela-tive sizes of prey consumed. The distribution of bodysizes of potential prey probably remains fairly constantwhen examined collectively across a time period of

245

Fig. 8. Trophic-niche slope (generated from ratio-based trophic-niche breadth vs predator size regressions) plotted against(a) maximum and (b) minimum prey size slopes (generated from prey size vs predator size regressions). Fitted regression equa-

tions and significance levels are indicated

Fig. 7. Average trophic-niche breadth plotted against averagepredator size for trophic niche breadths calculated using prey

size/predator size ratios. Error bars = ±1SE

Mar Ecol Prog Ser 208: 229–248, 2000

several years, such as the lifespan of a large predator.If this assumption is generally true, then the range ofrelative prey sizes available to a growing predator willnecessarily decrease with time. That is, the frequencydistribution of relative prey sizes will become skewedto the right as the frequency of larger prey size/preda-tor size ratios declines with increasing predator bodysize. In contrast to the results of previous research,Pepin & Penney (1997) observed increases in trophic-niche breadth for 6 of 11 species of larval fishes, whichled them to conclude that general size-based trendsmay not exist across taxa, but rather that patterns ofprey size use are species-specific. Our results highlightthe need for future research to examine ratio-basedtrophic-niche breadth across large size ranges andontogenetic stages to determine if generalizations canbe made across taxa.

The conclusions of previous ratio-based analysesthat prey size is a constant proportion of predator sizethrough ontogeny has recently been questioned(Houde 1997). For fishes, the range of absolute preysizes clearly increases as predators grow, regardless oflife stage. However, changes in relative prey sizes con-sumed are difficult to detect. When comparing pat-terns based on absolute prey sizes versus relative preysizes, the apparent differences in trophic-nichebreadths are not contradictory, but rather representdifferent perspectives of the predator-prey interaction.The expansion of absolute prey size range by mostpredators is probably representative of increases inbehavioral and morphological capability to captureand swallow large prey combined with high encounterrates and susceptiblity to capture of small prey. More-over, knowledge of the range of absolute prey sizesvulnerable to specific predators is critical in order toquantify the impact of predators on prey populationsand community structure. On the other hand, subtlechanges in the range of relative prey sizes eaten mayrepresent slight ontogenetic changes in predator for-aging abilities and/or ontogenetic shifts to prey typeswith different escape proficiencies. Further, modestchanges in ratio-based trophic-niche breadths may bespecific to particular life stages and may only bedetectable when analyzed on an appropriate scale. Forexample, Houde (1997) suggested that the foragingstrategy of growing larvae is one that ensures expan-sion of prey sizes in the diet in order to increaseencounter rates between larval fish and potential prey.

CONCLUSIONS

Asymmetric predator size - prey size distributions arefrequently observed across several animal taxa. Theconsistent inclusion of small-bodied prey in the diets of

large predators probably represents profitable forag-ing behaviors when size-dependent probabilities ofencounter and capture are combined with handlingcosts of prey. The prominent role of size-based rates ofencounter and capture in determining the types andsizes of prey consumed demonstrates the importanceof species specific morphological and behavioral char-acteristics of both predator and prey. Variation amongpredators in size-based feeding strategies can berelated to differences in predator gape allometry,foraging tactics, morphological specializations, andmigratory behavior as well as individual prey behaviorand morphology. How these species specific traitsscale ontogenetically between predator and prey willbe a major determinant of changes in resource use(prey size and type) by growing predators. Althoughthe range of absolute prey sizes eaten typicallyincreases with predator size, ontogenetic changes inrelative prey sizes eaten are often statistically unde-tectable, suggesting that changes in predator foragingabilities with size may be subtle and that their detec-tion requires analysis on an appropriate scale.

Acknowledgements. We gratefully acknowledge the staff ofthe Northeast Fisheries Science Center for collecting, editing,and organizing the food habits database and for allowing usto present it here. This research was supported by the Coop-erative Marine Education and Research program of theNational Oceanic and Atmospheric Administration and theHudson River Foundation for Science and EnvironmentalResearch. We also thank J. Boreman, J. Link, M. Sutherland,A. Stoner, and several anonymous referees for critical reviewsand comments that improved the manuscript. The US Gov-ernment is authorized to produce and distribute reprints forgovernmental purposes notwithstanding any copyright nota-tion that may appear hereon. The views expressed herein arethose of the authors and do not necessarily reflect the view ofNOAA or any of its sub-agencies.

LITERATURE CITED

Azarovitz TR (1981) A brief historical review of the WoodsHole laboratory trawl survey time series. Can Spec PublFish Aquat Sci 58:62–67

Beamish FWH (1978) Swimming capacity. In: Hoar WS, Ran-dall DJ (eds) Fish physiology. Academic Press, New York,p 101–187

Bigelow HB, Schroeder WC (1953) Fishes of the Gulf ofMaine. Fish Bull Fish Wildl Serv US 53:1–577

Blaxter JHS (1986) Development of sense organs and behav-ior of teleost larvae with special reference to feeding andpredator avoidance. Trans Am Fish Soc 115:98–114

Blaxter JHS, Fuiman LA (1990) The role of the sensory sys-tems of herring larvae in evading predatory fishes. J MarBiol Assoc UK 70:413–427

Breck JE (1993) Foraging theory and piscivorous fish: are for-age fish just big zooplankton? Trans Am Fish Soc 122:902–911

Brooking TE, Rudstam LG, Olson MH, VanDeValk AJ (1998)Size-dependent alewife predation on larval walleyes inlaboratory experiments. N Am J Fish Manage 18:960–965

246

Scharf et al.: Predator size - prey size relationships of fish predators

Brooks JL, Dodson SI (1965) Predation, body size, and compo-sition of plankton. Science 150:28–35

Buckel JA, Letcher BH, Conover DO (1998) Effects of adelayed onset of piscivory on the size of age-0 bluefish.Trans Am Fish Soc 127:576–587

Buijse AD, Houthuijzen RP (1992) Piscivory, growth, and size-selective mortality of age 0 pikeperch (Stizostedion luciop-erca). Can J Fish Aquat Sci 49:894–902

Cade BS, Terrell JW, Schroeder RL (1999) Estimating effectsof limiting factors with regression quantiles. Ecology 80:311–323

Charnov EL (1976) Optimal foraging: the marginal value the-orem. Theor Popul Biol 9:129–136

Christensen B (1996) Predator foraging capabilities and preyantipredator behaviours: pre-versus postcapture con-straints on size-dependent predator-prey interactions.Oikos 76:368–380

Edwards RL, Bowman RE (1979) Food consumed by continen-tal shelf fishes. In: Clepper H (ed) Predator-prey systemsin fisheries management. Sports Fishing Institute, Wash-ington, DC, p 387–406

Eggers DM (1977) The nature of prey selection by planktivo-rous fish. Ecology 58:46–59

Ellis T, Gibson RN (1997) Predation of 0-group flatfishes by0-group cod: handling times and size-selection. Mar EcolProg Ser 149:83–90

Folkvord A, Hunter JR (1986) Size-specific vulnerability ofnorthern anchovy, Engraulis mordax, larvae to predationby fishes. Fish Bull 84:859–869

Gaughan DJ, Potter IC (1997) Analysis of diet and feedingstrategies within an assemblage of estuarine larval fishand an objective assessment of dietary niche overlap. FishBull 95:722–731

Gerritsen J, Strickler JR (1977) Encounter probabilities andcommunity structure in zooplankton: a mathematicalmodel. J Fish Res Board Can 34:73–82

Gillen AL, Stein RA, Carline RF (1981) Predation by pellet-reared tiger muskellunge on minnows and bluegills inexperimental systems. Trans Am Fish Soc 110:197–209

Greene CH (1983) Selective predation in freshwater zoo-plankton communities. Int Rev Hydrobiol 68:297–315

Greene CH (1985) Planktivore functional groups and patternsof prey selection in pelagic communities. J Plankton Res 7:35–40

Greene CH (1986) Patterns of prey selection: implications ofpredator foraging tactics. Am Nat 128:824–839

Hall DJ, Cooper WE, Werner EE (1970) An experimental ap-proach to the production dynamics and structure of fresh-water animal communities. Limnol Oceanogr 15: 829–928

Hambright KD (1991) Experimental analysis of prey selectionby largemouth bass: role of predator mouth width andprey body depth. Trans Am Fish Soc 120:500–508

Harper DG, Blake RW (1988) Energetics of piscivorous preda-tor-prey interactions. J Theor Biol 134:59–76

Hart PJB, Hamrin SF (1990) The role of behaviour and mor-phology in the selection of prey by pike. In: Hughes RN(ed) Behavioural mechanisms of food selection. Springer-Verlag, Berlin, p 235–253

Houde ED (1997) Patterns and consequences of selective pro-cesses in teleost early life histories. In: Chambers RC, Trip-pel EA (eds) Early life history and recruitment in fish pop-ulations. Chapman & Hall, London, p 173–196

Hoyle JA, Keast A (1987) The effect of prey morphology andsize on handling time in a piscivore, the largemouth bass(Micropterus salmoides). Can J Zool 65:1972–1977

Hunter JR (1981) Feeding ecology and predation of marinelarvae. In: Lasker R (ed) Marine fish larvae: morphology,

ecology and relation to fisheries. University of WashingtonPress, Seattle, p 34–77

Ivlev VS (1961) Experimental ecology of the feeding of fishes.Yale University Press, New Haven

Juanes F (1994) What determines prey size selectivity in pis-civorous fishes? In: Stouder DJ, Fresh KL, Feller RJ (eds)Theory and application in fish feeding ecology. CarolinaUniversity Press, Columbia, p 79–100

Juanes F, Conover DO (1994) Piscivory and prey size selectionin young-of-the-year bluefish: predator preference or size-dependent capture success? Mar Ecol Prog Ser 114:59–69

Juanes F, Conover DO (1995) Size-structured piscivory:advection and the linkage between predator and preyrecruitment in young-of-the-year bluefish. Mar Ecol ProgSer 128:287–304

Keast A, Webb D (1966) Mouth and body form relative tofeeding ecology in the fish fauna of a small lake, LakeOpinicon, Ontario. J Fish Res Board Can 23:1845–1874

Keeley ER, Grant JWA (1997) Allometry of diet selectivity injuvenile Atlantic salmon (Salmo salar). Can J Fish AquatSci 54:1894–1902

Kerr SR (1974) Theory of size distribution in ecological com-munities. J Fish Res Board Can 31:1859–1862

Kislalioglu M, Gibson RN (1976) Prey ‘handling time’ and itsimportance in food selection by the 15-spined stickleback,Spinachia spinachia (L.). J Exp Mar Biol Ecol 25:151–158

Labropoulou M, Eleftheriou A (1997) The foraging ecology oftwo pairs of congeneric demersal fish species: importanceof morphological characteristics in prey selection. J FishBiol 50:324–340

Langton RW, Bowman RE (1980) Food of fifteen NorthwestAtlantic gadiform fishes. NOAA Natl Fish Serv Tech RepUS Dep Commerce SSRF-740:1–23

Langton RW, Bowman RE (1981) Food of eight NorthwestAtlantic pleuronectiform fishes. NOAA Natl Fish ServTech Rep US Dep Commerce SSRF-749:1–16

Litvak MK, Leggett WC (1992) Age and size-selective preda-tion on larval fishes: the bigger-is-better hypothesis revis-ited. Mar Ecol Prog Ser 81:13–24

Miller TJ, Crowder LB, Rice JA, Marschall EA (1988) Larvalsize and recruitment mechanisms in fishes: toward a con-ceptual framework. Can J Fish Aquat Sci 45:1657–1670

Mittelbach GG (1981) Foraging efficiency and body size: astudy of optimal diet and habitat use by bluegills. Ecology62:1370–1386

Mittelbach GG, Persson L (1998) The ontogeny of piscivoryand its ecological consequences. Can J Fish Aquat Sci55:1454–1465

Munk P (1992) Foraging behaviour and prey size spectra oflarval herring Clupea harengus. Mar Ecol Prog Ser 80:149–158

Munk P (1997) Prey size spectra and prey availability of larvaland small juvenile cod. J Fish Biol 51(Suppl A):340–351

Nielsen LA (1980) Effect of walleye (Stizostedion vitreum vit-reum) predation on juvenile mortality and recruitment ofyellow perch (Perca flavescens) in Oneida Lake, NewYork. Can J Fish Aquat Sci 37:11–19

Pearre S Jr (1986) Ratio-based trophic niche breadths of fish,the Sheldon spectrum, and the size-efficiency hypothesis.Mar Ecol Prog Ser 27:299–314

Pepin P, Penney RW (1997) Patterns of prey size and taxo-nomic composition in larval fish: are there general size-dependent models? J Fish Biol 51(Suppl A):84–100

Persson L (1990) Predicting ontogenetic niche shifts in thefield: what can be gained by foraging theory? In: HughesRN (ed) Behavioural mechanisms of food selection.Springer-Verlag, Berlin, p 303–321

247

Mar Ecol Prog Ser 208: 229–248, 2000

Persson L, Andersson J, Wahlstrom E, Eklov P (1996) Size-specific interactions in lake systems: predator gape limita-tion and prey growth rate and mortality. Ecology 77:900–911

Piet GJ, Pfisterer AB, Rijnsdorp AD (1998) On factors structur-ing the flatfish assemblage in the southern North Sea.J Sea Res 40:143–152

Platt T, Denman K (1978) The structure of pelagic marineecosystems. Rapp PV Réun Cons Int Explor Mer 173:60–65

Popova OA (1967) The ‘predator-prey’ relationship amongfish. In: Gerking SD (ed) The biological basis of freshwaterfish production. Blackwell Scientific Publications, Oxford,p 359–376

Popova OA (1978) The role of predaceous fish in ecosystems.In: Gerking SD (ed) Ecology of freshwater fish production.John Wiley & Sons, New York, p 215–249

Rice JA, Crowder LB, Rose KA (1993) Interactions betweensize-structured predator and prey populations: experi-mental test and model comparison. Trans Am Fish Soc122:481–491

Roughgarden J (1972) Evolution of niche width. Am Nat 106:683–718

Roughgarden J (1974) Niche width: biogeographic patternsamong Anolis lizard populations. Am Nat 108:429–442

Scharf FS, Buckel JA, Juanes F, Conover DO (1998a) Preda-tion by juvenile piscivorous bluefish (Pomatomus salta-trix): the influence of prey to predator size ratio and preytype on predator capture success and prey profitability.Can J Fish Aquat Sci 55:1695–1703

Scharf FS, Juanes F, Sutherland M (1998b) Inferring ecologi-cal relationships from the edges of scatter diagrams: com-parison of regression techniques. Ecology 79:448–460

Scharf FS, Yetter RM, Summers AP, Juanes F (1998c) Enhanc-ing diet analyses of piscivorous fishes in the NorthwestAtlantic through identification and reconstruction of origi-nal prey sizes from ingested remains. Fish Bull 96:575–588

Schoener TW (1971) Theory of feeding strategies. Annu RevEcol Syst 2:369–404

Schoener TW, Gorman GC (1968) Some niche differences in

three Lesser Antillean lizards of the genus Anolis. Ecology49:819–830

Sedberry, GR (1983) Food habits and trophic relationships ofa community of fishes on the outer continental shelf.NOAA Natl Fish Serv Tech Rep US Dep Commerce SSRF-773:1–56

Sheldon RW, Prakash A, Sutcliffe WH (1972) The size distrib-ution of particles in the ocean. Limnol Oceanogr 17:327–340

Shirota A (1970) Studies on the mouth size of fish in the larvaland fry stages. Bull Jpn Soc Sci Fish 36:353–368

Sih A, Moore RD (1990) Interacting effects of predator andprey behavior in determining diets. In: Hughes RN (ed)Behavioural mechanisms of food selection. Springer-Verlag, Berlin, p 771–796

Sokal RR, Rohlf FJ (1995) Biometry. The principles and prac-tice of statistics in biological research. WH Freeman andCompany, New York

StataCorp (1999) Stata statistical software: release 6.0. Col-lege Station, TX

Stein RA, Goodman CG, Marschall EA (1984) Using time andenergetic measures of cost in estimating prey value forfish predators. Ecology 65:702–715

Stephens DW, Krebs JR (1986) Foraging theory. PrincetonUniversity Press, Princeton

Wanzenböck J (1995) Changing handling times during feed-ing and consequences for prey size selection of 0+ zoo-planktivorous fish. Oecologia 104:372–378

Webb PW (1976) The effect of size on the fast-start perfor-mance of rainbow trout, Salmo gardneri, and a considera-tion of piscivorous predator-prey interactions. J Exp Biol65:157–177

Werner EE (1974) The fish size, prey size, handling time rela-tion in several sunfishes and some implications. J Fish ResBoard Can 31:1531–1536

Werner EE, Hall DJ (1974) Optimal foraging and the sizeselection of prey by the bluegill sunfish (Lepomismacrochirus). Ecology 55:1042–1052

Wilson DS (1975) The adequacy of body size as a niche differ-ence. Am Nat 109:769–784

248

Editorial responsibility: Howard Browman (ContributingEditor), Storebø, Norway

Submitted: January 3, 2000; Accepted: May 30, 2000Proofs received from author(s): November 13, 2000